marco/IsoMmo

Watch

Fork 0

Meta: map system (terrain + statics + editor + world) #115

New issue

Open

opened 2026-07-20 20:42:31 +02:00 by marco · 1 comment

marco commented

2026-07-20 20:42:31 +02:00

Owner

Meta / tracking issue for the map subsystem — how the game world is represented, authored, collided, and rendered. Groups the runtime, the tooling, the alpha arena, and the post-alpha world. Part of the Alpha epic (#107, pillar 2).

Pieces

✅ #89 — Static content system (runtime) [alpha, wave-1 — DONE, merged #144]: data model (dense terrain grid + sparse statics), server collision (blocks -> walkability + LOS), client render (viewport-culled), size-agnostic. The foundation everything else builds on.
✅ #113 — Map editor (minimal) [alpha, wave-2 — DONE, #147]: cross-platform MonoGame + ImGui tool to paint terrain + statics on the iso grid (pan/zoom, brush size, rectangle/fill, eraser, palette from tiledata). Authors maps visually.
#109 — PvP arena map [alpha, wave-2]: the alpha's playfield, authored with #113 on top of #89.
#114 — Large (~1000x1000) world map [post-alpha]: the real game world, authored once the tooling exists.

Dependency order

#89 (format + runtime) ✅ -> #113 (editor) ✅ -> #109 (arena authored with the editor) — next. #114 comes later.

Format decision (details in #89)

On-disk authoring format stays textual and diffable (two committed files: dense terrain grid + sparse statics list), read/written by the editor. Loader behind a swappable IMapSource / TileMap.Load(path) seam. Binary is deferred to world scale (#114) as a source->compiled just build-map step (mirroring the assets.isoa art pipeline), decided by measured need — never a hand-committed opaque blob.

Out of scope (future — open issues when needed)

Static height / z-layering, multi-tile statics, animated statics, map regions/zones metadata, in-game live editing.

No standalone Definition of Done: this meta closes when its alpha children (#89 ✅, #113 ✅, #109) are done; #114 tracks the post-alpha continuation.

Meta / tracking issue for the **map subsystem** — how the game world is represented, authored, collided, and rendered. Groups the runtime, the tooling, the alpha arena, and the post-alpha world. Part of the Alpha epic (#107, pillar 2). ## Pieces - ✅ **#89 — Static content system (runtime)** _[alpha, wave-1 — DONE, merged #144]_: data model (dense terrain grid + sparse statics), server collision (`blocks` -> walkability + LOS), client render (viewport-culled), size-agnostic. The foundation everything else builds on. - ✅ **#113 — Map editor (minimal)** _[alpha, wave-2 — DONE, #147]_: cross-platform MonoGame + ImGui tool to paint terrain + statics on the iso grid (pan/zoom, brush size, rectangle/fill, eraser, palette from tiledata). Authors maps visually. - **#109 — PvP arena map** _[alpha, wave-2]_: the alpha's playfield, authored with #113 on top of #89. - **#114 — Large (~1000x1000) world map** _[post-alpha]_: the real game world, authored once the tooling exists. ## Dependency order #89 (format + runtime) ✅ -> #113 (editor) ✅ -> **#109 (arena authored with the editor) — next**. #114 comes later. ## Format decision (details in #89) On-disk authoring format stays **textual and diffable** (two committed files: dense terrain grid + sparse statics list), read/written by the editor. Loader behind a swappable `IMapSource` / `TileMap.Load(path)` seam. Binary is deferred to world scale (#114) as a source->compiled `just build-map` step (mirroring the `assets.isoa` art pipeline), decided by measured need — never a hand-committed opaque blob. ## Out of scope (future — open issues when needed) Static height / z-layering, multi-tile statics, animated statics, map regions/zones metadata, in-game live editing. No standalone Definition of Done: this meta closes when its alpha children (#89 ✅, #113 ✅, #109) are done; #114 tracks the post-alpha continuation.

marco added the

epic

label

2026-07-20 20:42:31 +02:00

marco referenced this issue

2026-07-20 20:42:44 +02:00

Alpha — epic: PvP arena for friends #107

marco referenced this issue

2026-07-22 15:40:27 +02:00

refactor(world): data-driven tiledata behind an IMapFormat seam (#143) #145

marco referenced this issue from a commit

2026-07-22 15:44:13 +02:00

refactor(world): data-driven tiledata behind an IMapFormat seam (#143) (#145)

panda commented

2026-07-24 20:30:23 +02:00

Collaborator

Alpha milestone target: 132.2 million tiles

Owner's target for the alpha world (stated 2026-07-24). Not to be built now, but it should steer the decisions we take from here, because it invalidates two things we are doing today.


current map	128 x 96 = 12,288 tiles
alpha target	132,200,000 tiles (~11,500 x 11,500)
growth	~10,760x

What breaks at that scale

TextMapFormat is dead — one legend char per tile is a ~132 MB text file, read into a single string at load. This is exactly the swap the IMapFormat seam was introduced for (#143): a compiled/chunked binary becomes mandatory, not optional.
The dense TileMap grid is dead — 132 MB even at one byte per tile for a terrain id, and a TerrainDef per tile lands in the gigabytes. Needs chunking (load the chunks near players, drop the rest), a terrain palette, or both. This is the single biggest architectural consequence.
Authored static rows do not scale. Today's world.statics holds 291 rows for 12,288 tiles — 2.4% density. The same dressing at the target is ~3.1 million rows. Hand-authored placement cannot get there: the per-biome scatter (#140) has to be deterministic and generative (seed + biome rules, derived at load or on the fly), with authored rows reserved for landmarks. Worth deciding before we author much more by hand.

What does NOT break

AoI is radius-based, so per-player cost is unchanged by world size.
Persistence is unaffected — the map is static content, never in world.sav.
Statics as a sparse dict keyed by tile stays the right shape; it is the authoring of them that has to change.

Related: #143 (map format seam), #140 (per-biome scatter), #114 (author the large world).

### Alpha milestone target: **132.2 million tiles** Owner's target for the alpha world (stated 2026-07-24). Not to be built now, but it should steer the decisions we take from here, because it invalidates two things we are doing today. | | | |---|---| | current map | 128 x 96 = **12,288** tiles | | alpha target | **132,200,000** tiles (~11,500 x 11,500) | | growth | **~10,760x** | **What breaks at that scale** - **`TextMapFormat` is dead** — one legend char per tile is a ~132 MB text file, read into a single string at load. This is exactly the swap the `IMapFormat` seam was introduced for (#143): a compiled/chunked binary becomes mandatory, not optional. - **The dense `TileMap` grid is dead** — 132 MB even at one byte per tile for a terrain id, and a `TerrainDef` per tile lands in the gigabytes. Needs chunking (load the chunks near players, drop the rest), a terrain **palette**, or both. This is the single biggest architectural consequence. - **Authored static rows do not scale.** Today's `world.statics` holds 291 rows for 12,288 tiles — 2.4% density. The same dressing at the target is **~3.1 million rows**. Hand-authored placement cannot get there: the per-biome scatter (#140) has to be **deterministic and generative** (seed + biome rules, derived at load or on the fly), with authored rows reserved for landmarks. Worth deciding before we author much more by hand. **What does NOT break** - **AoI** is radius-based, so per-player cost is unchanged by world size. - **Persistence** is unaffected — the map is static content, never in `world.sav`. - **Statics as a sparse dict** keyed by tile stays the right shape; it is the *authoring* of them that has to change. Related: #143 (map format seam), #140 (per-biome scatter), #114 (author the large world).

panda added the

area:world

label

2026-08-08 15:49:53 +02:00

No milestone

No project

No assignees

2 participants

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

marco/IsoMmo#115

No description provided.

Rows
Columns